CN117129698A

CN117129698A - Abnormality alarm and retest system and method for full-automatic coagulation tester

Info

Publication number: CN117129698A
Application number: CN202311393807.1A
Authority: CN
Inventors: 杨军京
Original assignee: Beijing Zongci Technology Development Co ltd
Current assignee: Beijing Zongci Technology Development Co ltd
Priority date: 2023-10-26
Filing date: 2023-10-26
Publication date: 2023-11-28
Anticipated expiration: 2043-10-26
Also published as: CN117129698B

Abstract

The invention relates to the technical field of blood coagulation detection, in particular to an abnormality alarm and retest system and method of a full-automatic blood coagulation tester. Firstly, analyzing input real-time test data through a real-time abnormality detection module, and judging the state of a data signal to obtain an abnormal signal; then, verifying the abnormal signal, and confirming the abnormal signal or the false alarm state; sample isolation is performed based on the verified abnormal signal; based on the isolation status report and sample information to be retested, the purification and reset module starts to work; based on the purified and reset environmental state and the isolated sample, retesting is performed in a pure environment, and the system optimization and integration module integrates new test data; and finally, dynamically adjusting the energy use and processing speed of the whole machine according to the data and the operation load. The technical problems of inaccurate test and slower efficiency in the prior art during abnormal test are solved.

Description

Abnormality alarm and retest system and method for full-automatic coagulation tester

Technical Field

The invention relates to the technical field of blood coagulation detection, in particular to an abnormality alarm and retest system and method of a full-automatic blood coagulation tester.

Background

The coagulation test is an important clinical test for detecting the function and quantity of coagulation factors in blood. By this test, the coagulation status of the patient can be evaluated, and diagnosis of some diseases such as hemorrhagic diseases, thrombotic diseases, etc. can be made; the current blood coagulation test instrument may cause deviation of test results due to sample interference, operation errors or equipment faults, and the traditional alarm system may be limited to hardware faults, and is free from sample or operation abnormality, and when the test results are abnormal, manual intervention is needed for retesting, so that the efficiency is low; thus, there is a need to develop a system that automatically recognizes coagulation test anomalies, automatically triggers retests, and alarms in time.

There are many studies on coagulation tests, and Yang Junjing filed application numbers: "CN202010254634.5", publication date: 2020.06.23, name of invention: "a fully automatic coagulation tester", mainly comprising: the test platform comprises a bath temperature position for loading a test cup and heating, a first reagent position for storing an R1 reagent, a sample grade for storing a sample, a test position for carrying out cup test and a second reagent position for storing an R2 reagent; the sample adding arm unit is used for adding the R1 reagent and the sample into the test cup at the bath temperature position and adding the R2 reagent into the test cup at the test position; and the test cup conveying unit is used for conveying the test cup with the bath temperature position to the test position and discarding the tested test cup to the designated position. The full-automatic blood coagulation tester provided by the invention has the advantages of compact structure, excellent performance and high reliability.

However, the above technology has at least the following technical problems: the test is inaccurate and the efficiency is slower when the abnormality test is carried out.

Disclosure of Invention

The embodiment of the application solves the technical problems of inaccurate test and slower efficiency in the prior art when performing the abnormal test by providing the system and the method for the abnormal alarm and the retest of the full-automatic blood coagulation tester, and realizes the technical effects of the abnormal alarm and the retest of the full-automatic blood coagulation tester with high efficiency and high accuracy.

The application provides an abnormality alarm and retest system and method of a full-automatic coagulation tester, which specifically comprise the following technical scheme:

an abnormality alarm and retest system of a full-automatic coagulation tester, comprising:

the system comprises a real-time abnormality detection module, a false alarm filtering module, an automatic sample isolation and retesting module, a purifying and resetting module, a system optimizing and integrating module and a high-efficiency energy and speed management module;

when the real-time abnormality detection module detects abnormality, an abnormality signal is sent to the false alarm filtering module for false alarm filtering; once the false alarm filtering module confirms the abnormal signal, abnormal sample information is transmitted to the automatic sample isolation and retest module to carry out sample isolation and retest; the automatic sample isolation and retesting module interacts with the purification and reset module prior to retesting the sample; the system optimization and integration module is mutually connected with all other modules, receives the data flow and the instruction of each module, and then outputs optimized data processing and instructions; the operation of the system optimization integrated with the integration module will affect energy consumption and speed and interact with the high-efficiency energy and speed management module.

An anomaly alarm and retest system of a full-automatic coagulation tester, which further comprises:

the real-time abnormality detection module has the core functions of monitoring data change in the blood coagulation test process, and when the data exceeds a normal range, the data is considered as an abnormality, and corresponding abnormality signals are generated, wherein the abnormality signals comprise the types and the degrees of the abnormality;

the false alarm filtering module is used for deeply verifying the abnormal signals after receiving the abnormal signals generated by the real-time abnormal detection module, and confirming whether the abnormal signals are real abnormal or simply false alarm;

the automatic sample isolation and retest module immediately takes action after the false alarm filtering module confirms the abnormality, and isolates the sample confirmed to be abnormal; retesting the abnormal samples after isolation, but ensuring that the test environment is properly cleaned and reset before retesting;

the purification and reset module starts working when the automatic sample isolation and retesting system decides to carry out retesting, so that each retesting is ensured to be carried out in a pure test environment;

The system optimization and integration module is used for integrating the operation of all other modules and performing corresponding optimization, continuously receiving the data flow and the operation command of all other modules, and then outputting the optimized data processing mode and the module operation command;

the high-efficiency energy and speed management module is used for dynamically adjusting the energy use and the data processing speed of the whole machine and adjusting and optimizing the energy use and the processing speed of the instrument based on the operation load information received from the system optimization and integration module.

The abnormality alarming and retesting method for the full-automatic blood coagulation tester is applied to an abnormality alarming and retesting system of the full-automatic blood coagulation tester and comprises the following steps:

s1, analyzing input real-time test data through a real-time abnormality detection module, and judging the state of a data signal to obtain an abnormal signal;

s2, verifying the abnormal signal, and confirming the abnormal signal or the false alarm state;

s3, sample isolation is carried out based on the verified abnormal signals;

s4, based on the isolation state report and sample information to be retested, the purifying and resetting module starts to work;

s5, based on the purified and reset environment state and the isolated sample, re-testing in a pure environment, and integrating new test data by the system optimization and integration module;

S6, dynamically adjusting the energy use and processing speed of the whole machine according to the data and the operation load.

Preferably, the S1 specifically includes:

in the data correction of the earlier stage of data analysis, a nonlinear kernel-based correction algorithm is introduced.

Preferably, in the S1, the method further includes:

when data correction in the earlier stage of data analysis is performed, threshold truncation correction is introduced.

Preferably, in the S1, the method further includes:

when data analysis is performed on corrected data, minimum orthogonality factor decoupling is introduced.

Preferably, in the S1, the method further includes:

when data analysis is performed on the corrected data, an abnormal sensitivity adjustment algorithm based on density is further introduced.

Preferably, the S2 specifically includes:

in the verification, a method based on spectrum analysis is introduced: an anomaly verification filter; dynamic anomaly weight calculation is proposed.

Preferably, the S5 specifically includes:

during the test, an error correction coding system is started, which can capture and correct any transmission errors in real time; meanwhile, the data is subjected to preprocessing operation; calling a history database, and comparing the preprocessed data with the past data; and when data comparison is carried out, a high-latitude data consistency algorithm is introduced to quantify the consistency among high-dimensional data.

Preferably, in the step S5, the method further includes:

introducing a rapid model selection algorithm, and identifying a model most suitable for current data by adopting the rapid model selection algorithm.

The beneficial effects are that:

the technical schemes provided by the embodiment of the application have at least the following technical effects or advantages:

1. the application provides strong correction capability for data deviation by introducing nonlinear correction, thereby ensuring data quality and accuracy, and adopting threshold value cut-off correction to prevent data distortion and ensure that corrected data is still in an acceptable range; through decoupling of the minimum orthogonal factors, multiple collinearity is successfully eliminated, each coagulation index can be independently analyzed and interpreted, the density-based abnormal sensitivity adjustment algorithm ensures accurate identification of real abnormality, and misjudgment of normal values is avoided.

2. According to the application, the abnormal verification filter based on spectrum analysis is introduced, so that noise in data is effectively eliminated, abnormal signals are clearer and more definite, the possibility of false alarm is reduced, and the custom spectrum filter with a plurality of adjustable parameters is provided, so that the filter can be flexibly designed and adjusted according to actual requirements and characteristics of historical data, different verification requirements are met, and the severity of the abnormal condition can be more accurately evaluated by dynamically calculating the difference weight of the filtered data and the historical data, so that a powerful basis is provided for subsequent processing.

3. According to the application, the high latitude data consistency algorithm is introduced, so that the complexity of the data is reduced, and main information is not lost, so that the comparison and analysis of the data are more accurate, the consistency of new data and historical data is rapidly and accurately evaluated through the mahalanobis distance, and the timely discovery of abnormality or error is facilitated; meanwhile, the most suitable model is automatically selected for data analysis, so that time consumption and calculation amount are reduced.

4. The technical scheme of the application can effectively solve the technical problems of inaccurate test and slower efficiency when performing abnormal test, and provides strong correction capability for data deviation by introducing nonlinear correction, thereby ensuring the quality and accuracy of data, and adopting threshold value cut-off correction to prevent data distortion and ensure that corrected data is still in an acceptable range; the decoupling is performed through the minimum orthogonal factor, so that multiple collinearity is successfully eliminated, each coagulation index can be independently analyzed and interpreted, the density-based abnormal sensitivity adjustment algorithm ensures accurate identification of real abnormality, and misjudgment of normal values is avoided; the abnormal verification filter based on spectrum analysis is introduced, so that noise in data is effectively eliminated, abnormal signals are clearer and more definite, the possibility of false alarm is reduced, a custom spectrum filter with a plurality of adjustable parameters is provided, the filter can be flexibly designed and adjusted according to actual requirements and characteristics of historical data, different verification requirements are met, the severity of the abnormality can be more accurately estimated by dynamically calculating the difference weight of the filtered data and the historical data, and a powerful basis is provided for subsequent processing; and a high latitude data consistency algorithm is introduced, so that the complexity of data is reduced, and main information is not lost. The data comparison and analysis are more accurate, the consistency of the new data and the historical data is rapidly and accurately estimated through the mahalanobis distance, and the timely discovery of abnormality or error is facilitated; meanwhile, the most suitable model is automatically selected for data analysis, so that time consumption and calculation amount are reduced.

Drawings

FIG. 1 is a block diagram of an anomaly alarm and retest system of a fully automatic coagulation tester of the present application;

FIG. 2 is a flow chart of the anomaly alarm and retest method of the fully automatic coagulation tester of the present application;

Detailed Description

The embodiment of the application solves the technical problems of inaccurate test and slower efficiency in the prior art when performing the abnormality test by providing the abnormality alarm and retest system and method of the full-automatic blood coagulation tester, and the overall thinking is as follows:

the system comprises a real-time abnormality detection module, a false alarm filtering module, an automatic sample isolation and retesting module, a purifying and resetting module, a system optimizing and integrating module and a high-efficiency energy and speed management module; when the real-time abnormality detection module detects abnormality, an abnormality signal is sent to the false alarm filtering module for false alarm filtering; once the false alarm filtering module confirms the abnormal signal, the false alarm filtering module transmits abnormal sample information to the automatic sample isolation and retest module for sample isolation and retest; the automatic sample isolation and retesting module interacts with the purifying and resetting module before retesting the sample, so as to ensure that the testing environment is purified and reset correctly; the system optimization and integration module is mutually connected with all other modules, receives the data flow and the instruction of each module, and then outputs optimized data processing and instructions; the system optimization and integration module integration operations may affect energy consumption and speed, which may interact with the high-efficiency energy and speed management module to ensure that these operations are performed under optimal conditions; the nonlinear correction is introduced to provide a strong correction capability for data deviation, so that the data quality and accuracy are ensured, the threshold value cut-off correction is adopted to prevent data distortion, and the corrected data is still within an acceptable range; the decoupling is performed through the minimum orthogonal factor, so that multiple collinearity is successfully eliminated, each coagulation index can be independently analyzed and interpreted, the density-based abnormal sensitivity adjustment algorithm ensures accurate identification of real abnormality, and misjudgment of normal values is avoided; the abnormal verification filter based on spectrum analysis is introduced, so that noise in data is effectively eliminated, abnormal signals are clearer and more definite, the possibility of false alarm is reduced, a custom spectrum filter with a plurality of adjustable parameters is provided, the filter can be flexibly designed and adjusted according to actual requirements and characteristics of historical data, different verification requirements are met, the severity of the abnormality can be more accurately estimated by dynamically calculating the difference weight of the filtered data and the historical data, and a powerful basis is provided for subsequent processing; and a high latitude data consistency algorithm is introduced, so that the complexity of data is reduced, and main information is not lost. The data comparison and analysis are more accurate, the consistency of the new data and the historical data is rapidly and accurately estimated through the mahalanobis distance, and the timely discovery of abnormality or error is facilitated; meanwhile, the most suitable model is automatically selected for data analysis, so that time consumption and calculation amount are reduced.

In order to better understand the above technical solutions, the following detailed description will refer to the accompanying drawings and specific embodiments.

Referring to fig. 1, the abnormality alarm and retest system of the fully automatic coagulation tester according to the present application comprises the following parts:

the real-time abnormality detection module has the core functions of monitoring data change in the blood coagulation test process, and when the data exceeds a normal range, the data is considered to be an abnormality and a corresponding abnormality signal is generated, wherein the abnormality signal comprises the type and the degree of the abnormality;

the false alarm filtering module starts to act after receiving the abnormal signals generated by the real-time abnormal detection module, and the false alarm filtering module deeply verifies the abnormal signals to determine whether the abnormal signals are real abnormal or simply false alarm;

the automatic sample isolation and retest module takes action immediately after the error report filtering module confirms the abnormality, and isolates the sample confirmed to be abnormal so as to ensure that other tests are not influenced; after isolation, a proper time is selected to retest the samples, but before retesting, it is ensured that the test environment is properly cleaned and reset;

The purification and reset module starts to work when the automatic sample isolation and retesting system decides to carry out retesting, so that each retesting is ensured to be carried out in a pure testing environment;

the system optimizing and integrating module integrates the operation of all other modules and performs corresponding optimization to ensure the efficient operation of the whole instrument, continuously receives the data flow and the operation command of all other modules, and then outputs the optimized data processing mode and the module operation command.

The high-efficiency energy and speed management module is used for dynamically adjusting the energy use and the data processing speed of the whole machine in order to keep the blood coagulation tester running efficiently and stably, and can adjust and optimize the energy use and the processing speed of the instrument based on the operation load information received from the system optimization and integration module so as to ensure that all operations are executed under the optimal condition;

the connection between the modules is as follows:

when the real-time abnormality detection module detects abnormality, an abnormality signal is sent to the false alarm filtering module for false alarm filtering;

once the false alarm filtering module confirms the abnormal signal, the false alarm filtering module transmits abnormal sample information to the automatic sample isolation and retest module for sample isolation and retest;

The automatic sample isolation and retesting module interacts with the purifying and resetting module before retesting the sample, so as to ensure that the testing environment is purified and reset correctly;

the system optimization and integration module is mutually connected with all other modules, receives the data flow and the instruction of each module, and then outputs optimized data processing and instructions;

the system optimization and integration module integration operations may affect energy consumption and speed, which may interact with the high-efficiency energy and speed management module to ensure that these operations are performed under optimal conditions.

Referring to fig. 2, the abnormality alarm and retest method of the full-automatic coagulation tester according to the present application comprises the steps of:

the acquired biochemical signals of the blood coagulation testing instrument are converted into analog signals by using a high-precision sensor, then the analog signals are converted by an analog-digital converter (ADC) to generate digital data, standardized digital blood coagulation testing data are obtained, and the digital data are transmitted to a processing unit through a communication protocol (such as USB, bluetooth and the like) to perform data preprocessing; the data preprocessing utilizes a sliding window filter to reduce data noise, and the data is processed in a standardized way by using a Z-score method; the coagulation test data may be subject to errors due to daily calibration of the device, operator differences, even characteristics of the patient sample, etc. Therefore, further correction of the data is required:

In order to solve the nonlinear deviation of data caused by the change of time and environment, the application introduces a nonlinear kernel-based correction algorithm, wherein the nonlinear kernel-based correction algorithm is specifically realized as follows:

，

this is a cubic polynomial function, in which,representing raw, uncorrected coagulation test values; a. b,c and d are undetermined coefficients whose values are determined by using a least square method to minimize the difference between the original data and the correction data; obtaining corrected data->；

Further, to avoid that under certain extreme conditions, nonlinear correction leads to data distortion, the present application introduces threshold truncation correction; the method comprises the following steps:

，

in order to prevent data distortion, when the correction value exceeds a certain range, it is truncated to the upper or lower limit of the range,and->Is a preset upper limit threshold and a preset lower limit threshold, and is obtained according to an empirical method;

finally, final correction data are obtained and stored for later use;

further, data analysis and comparison are performed on the corrected data:

before any complex analysis is performed, descriptive statistics are firstly required for the data to obtain an overview of the data, the descriptive statistics not only provide rough distribution and central trend of the data, but also establish a basis for subsequent analysis, and parameters and initial thresholds are set; calculating the average value, standard deviation, median and quartile range of each coagulation index; identifying any possible outliers and outliers;

Through the process, a map of the original data is obtained, which is helpful for identifying possible data problems, and provides reference indexes and preliminary abnormal identifications for subsequent minimum orthogonal factor decoupling and density-based abnormal sensitivity adjustment algorithms;

to solve the problem of multiple collinearity caused by highly correlated coagulation indexes, a minimum orthogonal factor decoupling is introduced, the minimumDecoupling by small orthogonality factor, firstly regarding corrected blood coagulation test data as matrixWherein each column represents a coagulation index, calculating +.>Covariance matrix>：

，

Re-pair covariance matrixAnd (3) performing characteristic decomposition:

，

wherein,is a characteristic value matrix; />Is a feature vector matrix;

by characteristic valueTo construct a decoupling matrix +.>：

，

Calculating decoupled data：

，

Outliers identified in descriptive statistics are particularly important in this step because they may have an impact on covariance matrices and feature decomposition, and furthermore, by comparing descriptive statistics of data with decoupled results, the decoupling process can be further optimized;

further, in order to avoid high sensitivity of outliers, a density-based outlier sensitivity adjustment algorithm is introduced, which first calculates data after a minimum orthogonality factor decoupling process Local Density of each dot of (3)>：

，

Wherein,is data point +.>Is a local density of (c). A high value means that there are more data points around the point, while a low value means that the point may be abnormal or noisy, is local density +.>Is an element of any one of the following; />Is a width parameter of the gaussian kernel, which can be set according to the data distribution range in descriptive statistics; />Is the total number of data points in the dataset;

reuse of local densityDysregulated sensitivity->：

，

Wherein,representing a local density maximum; />Is the abnormal sensitivity of data point i, is the abnormal sensitivity +.>Any one element of (a) and (b);

finally, use sensitivityUpdating the anomaly score:

，

wherein,the method is an original anomaly score obtained according to the prior anomaly detection algorithm Isolation Forest under the condition of not considering the data density;

next, classifying the data points using an anomaly score derived from a density-based anomaly sensitivity adjustment algorithm, i.e., determining a threshold beyond which the data points are considered anomalies, which may be set based on a quartile range in descriptive statistics; and labeling each data point to indicate whether it is normal or abnormal;

Obtaining an abnormal signal through the process;

the application provides strong correction capability for data deviation by introducing nonlinear correction, thereby ensuring data quality and accuracy, and adopting threshold value cut-off correction to prevent data distortion and ensure that corrected data is still in an acceptable range; through decoupling of the minimum orthogonal factors, multiple collinearity is successfully eliminated, each coagulation index can be independently analyzed and interpreted, the density-based abnormal sensitivity adjustment algorithm ensures accurate identification of real abnormality, and misjudgment of normal values is avoided.

in the abnormal signal verification link, firstly, through an abnormal signal mark, automatically searching and comparing the recent similar abnormal data in a database by using the mark to obtain the history data reference of the recent similar abnormal, and at the same time, after detecting the abnormal signal, rapidly starting continuous sampling for a plurality of times to capture the data around the abnormality, so as to ensure accurate judgment on the continuity of the abnormality and generate a continuous temporary data sequence;

next, the sequences are combined with historical data extracted from the database for in-depth analytical comparisons:

First, it is critical to adjust the continuous temporary data to a common baseline, which can be accomplished by the following equation:

，

wherein,is the adjusted continuous temporary data, +.>Is the current continuous temporary dataIs the average of the historical data. In this way the first and second light sources,thus, the temporary data and the historical data are compared;

however, when verification is performed, temporary data is affected by environmental noise, so that verification is inaccurate, and in order to solve the problem, the application introduces a method based on spectrum analysis: anomaly verification filter:

first, the data is converted into the frequency domain:

，

wherein,the frequency is the corresponding data frequency and represents the speed of various signal changes in the data;

then, a custom spectral filter is usedTo eliminate the noise part in the data, the design of the filter needs to screen out the main useful frequency components and weaken the noise part based on the spectral characteristics of the historical data:

，

wherein,the self-defined spectrum filter is designed to remove noise which may cause false alarm; />Is calculated according to the formula:

，

wherein,current frequency, fourier transform and sampling rate from the signal; / >Is the maximum frequency of the signal, from the fourier transform and the sampling rate of the signal; />The method is to adjust the coefficient of window amplitude, control the window amplitude, and further influence the attenuation degree of the window on each part of the signal, and set by a designer of the system according to the required filter property; />Adjusting the coefficient of the depth of the sinusoidal modulation, controlling the depth of the sinusoidal modulation, playing a role in adjusting the frequency response of the filter, set by the designer of the system according to the desired filter properties; />The central frequency of sinusoidal modulation is selected, which part of the signal is regulated is determined, and the signal frequency band to be regulated is set according to the requirement; />Is an attenuation coefficient, influences the response of the filter to the high-frequency component, increases an exponential attenuation for the filter, and the response to the high-frequency component can be attenuated more rapidly, and is set by a system designer according to the requirement of the high-frequency response;

finally, the inverse fast fourier transform is performed to convert the data back to the time domain:

，

the selection of the filter is based on the analysis of the spectral characteristics of the historical data, and the corresponding main frequency part is selected; finally, data processed by an anomaly verification filter is obtained ；

Next, considering how to compare with the history data and confirm the anomaly, the present application proposes "dynamic anomaly weight calculation":

first, a weight is calculated, the weight being based on the difference of the filtered data from the historical data:

，

next, a threshold value is setIf->Beyond this threshold, an anomaly is considered to occur, which is determined based on the variance of the historical data; obtaining abnormality determination weight->；

Combining the two algorithms, verifying the abnormal data and further determining the abnormal data state;

after the deep analysis and comparison processing, a more accurate judging result of the abnormal signal is obtained, whether the abnormal signal is confirmed or the abnormal signal is in a false report state;

and finally, determining the next action according to the comparison result, and if the comparison result is confirmed to be real abnormality after careful check, automatically triggering an isolation mechanism to ensure that the problem data signal is isolated and carrying out subsequent processing. Otherwise, if the analysis determines that the error report is generated, the system can reenter the normal test flow to continue the detection of other samples;

according to the application, the abnormal verification filter based on spectrum analysis is introduced, so that noise in data is effectively eliminated, abnormal signals are clearer and more definite, the possibility of false alarm is reduced, and the custom spectrum filter with a plurality of adjustable parameters is provided, so that the filter can be flexibly designed and adjusted according to actual requirements and characteristics of historical data, different verification requirements are met, and the severity of the abnormal condition can be more accurately evaluated by dynamically calculating the difference weight of the filtered data and the historical data, so that a powerful basis is provided for subsequent processing.

S3, sample isolation is carried out based on the verified abnormal signals;

in an anomaly alarm and retest system of a full-automatic coagulation tester, sample isolation is an important link, when the system receives an validated anomaly signal, the anomaly signal is analyzed through a data analysis algorithm, the batch number, the sample position, the anomaly type and the like of an anomaly sample are identified, further, risk assessment is carried out on the analyzed anomaly sample information through the system, for example, the probability that other samples are possibly influenced is calculated, and the number and the position of required isolation containers are estimated by using a preset decision tree or a machine learning model;

the control unit then queries the database or memory to check the number of isolation containers currently available. If the isolation container is not enough, the system can send out an emergency alarm through the internal communication module, and manual intervention is required; however, if the isolation container is sufficient, the scheduling algorithm will calculate the optimal container allocation policy for the outlier samples. With these preparations, a precision robotic arm within the system can be activated, utilizing a PID control algorithm to ensure that the sample is stably moved from the home position to the isolation container. In the moving process, the system can monitor physical conditions such as temperature, pressure and the like of the sample in real time, so that more problems cannot be caused due to environmental changes;

Then, once the sample is successfully isolated to the container, a high-resolution camera and an image recognition algorithm of the system are started, the successful isolation of the sample is confirmed, and the sample in the isolated container is identified and recorded by utilizing an RFID (radio frequency identification) or two-dimensional code scanning technology;

finally, in order to record the whole process and provide data for the subsequent steps, the information integration module integrates all data into a database, and meanwhile, the report generation module generates a detailed isolation state report according to the information in the database and automatically sends the report to a preset receiving end through an SMTP protocol or an API interface so as to ensure real-time monitoring and data acquisition at any time;

firstly, carrying out state inspection and evaluation through a built-in sensor, capturing environmental parameters such as pH, temperature and humidity in the instrument, and comparing the captured environmental parameters with a preset normal range to determine an abnormal part to form an evaluation report, wherein the evaluation report indicates instrument parts and areas need to be purified and reset;

next, based on this assessment report, dynamically generating an equipment cleaning strategy using a decision tree algorithm to ensure that each of the components involved and the cleaning method to be used are enumerated in detail, and then selecting and dispensing the appropriate cleaning agents and tools using an automated warehouse management system, while the robotic arm prepares the required cleaning materials according to the previously formed cleaning plan;

After the preparation work is finished, the unmanned mechanical arm starts to operate according to a cleaning plan, the equipment is cleaned locally or wholly, and after the cleaning is finished, the ultraviolet lamp is used for disinfecting the equipment for a short time, so that the test environment is aseptic. After this stage is completed, a cleaned and purified equipment status report is formed;

then, the control system checks all parameters in the instrument according to the status report, compares the parameters with preset standard configuration, resets any parameters which do not accord with the preset parameters, ensures that the instrument reaches a completely reset state and is completely ready for retesting;

finally, to ensure that all steps are completed correctly, the instrument starts a final inspection and verification program, the program uses the built-in sensor and the camera to perform comprehensive inspection, and simultaneously operates a simple self-detection to ensure that the whole system is error-free and ready for coagulation test;

firstly, detecting temperature, humidity and other key parameters of a test area in real time through a sensor, evaluating the suitability of the environment, and then automatically adjusting the instrument to an initial state by using embedded control software, such as cleaning a probe, resetting a suction head and the like; in order to ensure the accuracy, the system starts temperature control equipment, ensures that the test area reaches the standard operation temperature, and starts an environment stabilization system, so that the temperature is stable and is not interfered by external factors;

Further, the high-resolution camera scans the appearance of the sample, judges the integrity of the sample, and once the integrity of the sample is confirmed, the accurate mechanical arm can start and move the sample from the isolation area to the test area according to a preset track; to ensure fluency in subsequent operations, infrared positioning techniques are used to accurately position and pose the sample;

on a Graphical User Interface (GUI), a test program for a particular sample is selected. Subsequently, the AI assistant automatically analyzes the sample information, recommends optimal test parameters and adjusts accordingly. In order to ensure the transparency of the whole testing process, the system starts the video monitoring of the testing process and simultaneously displays the data stream in real time;

further, to ensure the integrity and accuracy of the data, an error correction coding system is started during the test, which can capture and correct any transmission errors in real time; meanwhile, the data is subjected to preprocessing operations such as digital filtering, signal amplification, baseline adjustment and the like; in order to ensure data security, the processed data is written into a cache database in real time and backed up to a hard disk for long-term storage;

next, a deep learning algorithm is used for feature extraction and anomaly detection of new data. The historical database is called, new data is compared with the past data, and consistency and reliability of the new data are evaluated. To analyze and understand the data more deeply, the system also uses statistical tools to perform regression analysis, principal component analysis, etc., from which key parameters are extracted.

Next, the processed test data is subjected to data integration processing, firstly, in order to ensure the integrity and accuracy of the data, a system accesses historical data from a database or cloud storage, screens out corresponding historical data according to the type and date of a current test sample, and loads the corresponding historical data into a current working space;

next, using data comparison algorithms, such as correlation analysis, mean square error calculation, etc., to compare new and old data, in this step, the comparison of key indicators is particularly important, since they ensure that there is no significant deviation or anomaly, if any anomaly is detected at this stage, an anomaly report will be generated and the subsequent further analysis will be entered;

when data comparison is carried out, high dimensionality and high complexity of the data exist, so that the traditional correlation analysis and mean square error calculation cannot accurately reflect the real relation of the dataReducing the dimensionality of the data to +.>Wherein->（/>To the original dimension):

，

wherein,the data matrix after dimension reduction still contains main information of original data, but the dimension of the data matrix is reduced;

Calculating the average value of the data after dimension reductionAnd covariance matrix->：

，

Wherein,the average value of the data after the dimension reduction is represented as a data center in a dimension reduction space; />Is covariance matrix of the data after dimension reduction, describing the relationship between the data characteristics after dimension reduction;

re-calculating the mahalanobis distance of the new data and the history data in the dimension-reduced space：

，/>

Wherein,is a dimension-reduced representation of new data, representing the position of new samples in the same dimension-reduced space;

the algorithm can effectively quantify the consistency among high-dimensional data, reduce noise and unnecessary information, and more accurately reflect the internal relation of the data;

further, in the data integration process of the full-automatic coagulation tester, the mahalanobis distanceCan be used for evaluating the consistency between new data and historical data, and specifically comprises the following steps: first, a threshold value of the Mahalanobis distance is set empirically>This threshold is determined according to the expert's experience and the calculated Mahalanobis distance +.>And threshold->Compare if->The fact that the new data and the historical data have large differences possibly caused by abnormal values or errors in the testing process is indicated, and an alarm mechanism is triggered at the moment to remind relevant personnel to check; if- >The new data is more consistent with the historical data, and the next operation can be carried out;

in the case of no anomaly, statistical analysis of the data can be further performed by using statistical tools, such as calculating the mean, standard deviation, confidence interval, etc. of the data, and in order to better understand complex data patterns or trends, it is also important to select a suitable data model for fitting, such as linear regression, logistic regression, or neural network, which not only helps to analyze the current data, but also can predict future data trends or analyze potential factors;

the present application introduces a fast model selection algorithm that selects a set of candidate modelsTraining data->As model input, when algorithm implementation is performed, taking the mth model as an example, first, initializing model evaluation score +.>The method comprises the steps of carrying out a first treatment on the surface of the Then try each model one by one using gradient lifting means +.>And prediction error of model output +.>And (3) optimizing:

，

wherein,is a model->Is used for the prediction error of (a). Measuring the difference between the model prediction and the actual observation; />A true observed value, an original result obtained directly from the coagulation tester; / >Is a predictive value of a model, model->Based on input data->Is a predicted result of (a);

further, update model scores：

，

Wherein,is a model->Is the model +.>Performance index in the current iteration; />Is learning rate, a positive constant, used to adjust the speed of model score update; />Representing an initial score;

further, selecting the model with the lowest score as the optimal model to obtain the optimal model；

The algorithm greatly reduces the calculation amount and time required by model selection, and can accurately identify the model most suitable for the current data;

however, if an anomaly is detected in the previous comparison, this stage requires deep anomaly analysis, which involves deep mining of the cause of the anomaly, which may include instrument failure, sample problems or other external factors, and data visualization tools play a key role here as well, which show the anomaly data in detail through various graphs such as scatter plots, bin patterns, etc., and based on the result of the anomaly analysis, give corresponding advice such as whether to re-test or repair a certain part of hardware;

finally, all of these reports and analyses will be integrated into one complete test data integration report. The report details the testing process, method, results and advice, and is formatted and optimized to ensure the readability and accuracy;

Further, the system intelligently recommends a subsequent processing mode of the sample according to the test result, such as storage, discarding or further analysis, the accurate mechanical arm can move the sample to a corresponding position according to the suggestions, and finally, the system can integrate all data and images to generate a complete and visual test report, and simultaneously, the functions of downloading, printing and the like are provided;

finally, retested result data are obtained;

the application reduces the complexity of data by introducing a high latitude data consistency algorithm, and does not lose main information. The data comparison and analysis are more accurate, the consistency of the new data and the historical data is rapidly and accurately estimated through the mahalanobis distance, and the timely discovery of abnormality or error is facilitated; meanwhile, the most suitable model is automatically selected for data analysis, so that time consumption and calculation amount are reduced.

Firstly, data acquisition is carried out by using a sensor and a built-in analysis tool, various sensor data on equipment are read in real time through a multi-channel input/output (IO) interface, time series analysis is adopted to classify and mark the data, and marked working state data is generated;

Next, in order to predict energy consumption, an energy consumption model is built, using a machine learning algorithm, such as a decision tree, which predicts the upcoming energy demand based on past data, which continuously adjusts itself to accommodate new device states during the training of the model, thereby outputting predicted values of energy demand;

in terms of speed, the processing speed requirement is judged through a built-in timer and a processor state reading tool, and according to the retested data complexity and the length of the equipment task queue, the system can judge the processing speed requirement in real time and determine whether a higher or lower processing speed is required or not by using a corresponding algorithm, so that the evaluation of the processing speed requirement is obtained;

along with the prediction of the energy demand, the equipment can dynamically adjust the output of a battery or a power supply, and a fuzzy logic controller is used, so that the power supply voltage and current of the power supply can be dynamically adjusted according to the predicted value of the energy demand, and the real-time energy supply state is ensured to be kept optimal;

the processing speed can also be adjusted according to the requirements by using hardware acceleration or dynamic frequency adjustment technology, the clock frequency of the processor can be dynamically changed or certain cores can be started/shut down according to the evaluation of the processing speed requirements, and for non-critical tasks, a low-power consumption mode or background processing can be considered to ensure that the device always operates in the optimal speed and mode;

Finally, in order to ensure the stability and continuous optimization of the device, the running state of the device must be continuously collected and fed back, the running state of the device is collected in real time by using the self-detection function and the internal state monitoring tool of the device, if any abnormality is detected, the system immediately triggers an alarm and starts a self-correction program, thereby ensuring the stable running of the device and generating a real-time device state monitoring report and an abnormality record;

in conclusion, the abnormality alarming and retesting system and method of the full-automatic coagulation tester are completed.

The technical scheme provided by the embodiment of the application at least has the following technical effects or advantages:

Effect investigation:

the technical scheme of the application can effectively solve the technical problems of inaccurate test and slower efficiency when the abnormal test is carried out, and the system or the method provides strong correction capability for data deviation by introducing nonlinear correction through a series of effect researches, thereby ensuring the data quality and accuracy, adopting threshold value cut-off correction, preventing data distortion and ensuring that corrected data is still in an acceptable range; the decoupling is performed through the minimum orthogonal factor, so that multiple collinearity is successfully eliminated, each coagulation index can be independently analyzed and interpreted, the density-based abnormal sensitivity adjustment algorithm ensures accurate identification of real abnormality, and misjudgment of normal values is avoided; the abnormal verification filter based on spectrum analysis is introduced, so that noise in data is effectively eliminated, abnormal signals are clearer and more definite, the possibility of false alarm is reduced, a custom spectrum filter with a plurality of adjustable parameters is provided, the filter can be flexibly designed and adjusted according to actual requirements and characteristics of historical data, different verification requirements are met, the severity of the abnormality can be more accurately estimated by dynamically calculating the difference weight of the filtered data and the historical data, and a powerful basis is provided for subsequent processing; and a high latitude data consistency algorithm is introduced, so that the complexity of data is reduced, and main information is not lost. The data comparison and analysis are more accurate, the consistency of the new data and the historical data is rapidly and accurately estimated through the mahalanobis distance, and the timely discovery of abnormality or error is facilitated; meanwhile, the most suitable model is automatically selected for data analysis, so that time consumption and calculation amount are reduced.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.

It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.

Claims

1. The abnormality alarm and retest system of the full-automatic blood coagulation tester, characterized by comprising:

2. The anomaly alarm and retest system of a fully automatic coagulation test instrument as claimed in claim 1, further comprising:

3. The abnormality alarm and retest method of the full-automatic blood coagulation tester, which is applied to the abnormality alarm and retest system of the full-automatic blood coagulation tester as claimed in claim 1, is characterized by comprising the following steps:

s3, sample isolation is carried out based on the verified abnormal signals;

4. The method for anomaly alarm and retest of full-automatic coagulation test apparatus according to claim 3, wherein the step S1 specifically comprises:

5. The method for anomaly alarm and retest of full-automatic coagulation test apparatus as claimed in claim 4, wherein in the step S1, further comprising:

6. The method for anomaly alarm and retest of full-automatic coagulation test apparatus as claimed in claim 5, wherein in S1, further comprising:

7. The method for anomaly alarm and retest of full-automatic coagulation test apparatus as claimed in claim 6, wherein in S1, further comprising:

8. The method for anomaly alarm and retest of full-automatic coagulation test apparatus according to claim 3, wherein S2 specifically comprises:

9. The method for anomaly alarm and retest of full-automatic coagulation test apparatus according to claim 3, wherein S5 specifically comprises:

10. The method for anomaly alarm and retest of full-automatic coagulation test apparatus as claimed in claim 9, further comprising, in S5: